|
Abstract:
|
Early detection of bladder cancer could be life saving but the consequences of false positives and false negatives bear their own costs. Recent advances in genomics and proteomics allow biomedical researchers to use biomarkers measured by noninvasive and economic procedures (e.g., samples from serum, urine, or stool) for disease classification. Currently, more than 20 known urine-based biomarkers have been identified. Often one maker alone is not sufficient enough for cancer diagnosis. A panel of biomarkers could be used to improve diagnosis accuracy. To evaluate various bladder cancer diagnostic systems, we propose using a combining sensitivity and specificity of heterogeneous biomarkers through likelihood ratio decision criterion taking into account the number of individual tests and correlations between the tests to quantify various system performances.
|
